Ibm Big Data Whitepaper
-
Upload
anonymous-pkvcsg -
Category
Documents
-
view
235 -
download
0
Transcript of Ibm Big Data Whitepaper
-
8/15/2019 Ibm Big Data Whitepaper
1/13
White Paper
A Visualization is Worth aThousand Tables: How IBMBusiness Analytics LetsUsers See Big Data
Contents
Executive Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
Too much data, not enough information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
Only statisticians like tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
Spreadsheets do not equal data visualization tools . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
The Solution: Rich Visualization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 What is visualization? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
Why visualization is different than reporting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
Exploratory data analysis and predictive analytics – Visually . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
How analytics visualization creates competitive advantages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .12
IBM Visualization: Natural, Intuitive, and Powerful . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
Visualizations that automatically suit the data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
Interactivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
Extensibility and ready-made solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
Conclusion: Visualization Makes Big Data Accessible . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
®
®
About Ziff Davis B2B
Ziff Davis B2B is a leading provider of research to technology buyers and high-quality
leads to IT vendors. As part of the Ziff Davis family, Ziff Davis B2B has access to over
50 million in-market technology buyers every month and supports the company’s core
mission of enabling technology buyers to make more informed business decisions.
Contact Ziff Davis B2B
100 California Street, Suite 650
San Francisco, CA 94111
Tel: 415.318.7200 | Fax: 415.318.7219
Email: [email protected]
www.ziffdavis.comCopyright © 2014 Ziff Davis B2B. All rights reserved.
-
8/15/2019 Ibm Big Data Whitepaper
2/13
ziffdavis.com 2 of 13
Ziff Davis | White Paper | A Visualization is Worth a Thousand Tables
ziffdavis.com
Executive Summary
Too often, organizations must resort to presenting data in summary tables and generating
graphs and charts with spreadsheet applications. As business data gives way to big data, these
approaches are increasingly inadequate and fail to produce useful, accessible views of data.
At the same time, an increasing number of users have the desire, wherewithal, and need to
engage in their own exploratory data analysis projects beyond reviewing standard dashboards
and scorecards.
In both cases, visualization tools are often better suited than spreadsheets and tables to
helping stakeholders truly understand their data. IBM has several solutions that let technicaland non-technical users alike dispense with ineffective, legacy approaches and intuitively
explore and visualize data, regardless of its complexity. This paper explores the advantages of
visualization for data exploration and analysis as well as specific tools from IBM that support
analytics across an organization.
http://www.ziffdavis.com/http://www.ziffdavis.com/http://www.ziffdavis.com/http://../jramires/Library/Caches/Adobe%20InDesign/Version%207.5/en_US/InDesign%20ClipboardScrap1.pdf
-
8/15/2019 Ibm Big Data Whitepaper
3/13
ziffdavis.com 3 of 13
Ziff Davis | White Paper | A Visualization is Worth a Thousand Tables
Introduction
Organizations are expected to be “data-driven.”
They collect metrics, store user preferences, track
customer behavior, and collect machine data by
the terabyte. But as the amount of data a business
collects increases, often its ability to act on the data
decreases. The volume and complexity of the data
are simply too overwhelming and many organizations
struggle to see the forest for the trees. The predictive
value of the data and the competitive advantages it
can create are lost in the noise.
This problem is compounded as lines of business
require interactive access to data and incorporate
business analytics in daily decision-making. What was
once the sole responsibility of statisticians and data
scientists has trickled down to a much broader group
of business users. Although these users may lack
statistical expertise, their deep understanding of the
business itself can make access to the right analytical
tools invaluable in improving operational efficiencies
and identifying new opportunities. In mid-sized
businesses, where there often aren’t resources for
dedicated data scientists, these tools are even more
critical.
Too much data, not enough information
Virtually all mid- and large-sized organizations are
now contending with big data. Whether the data is big
because of sheer volume, the velocity of its collection,
or the complexity of the data itself, many businesseshave reached a point where traditional means of
analysis and reporting simply aren’t adequate to
derive real predictive insights from the data.
While we have well-defined practices, software, and
hardware for collecting and storing data ranging from
high-speed transactional processing to unstructured
The evolving role of the
data scientist
Many large organizations have
employed statisticians for years
to analyze data, identify trends,
support strategic planning,
and improve processes.
More recently, the term “data
scientist” has started appearing
in corporate job listings, as
business analytics have become
more critical to success and
competitiveness. Though often
statisticians by trade, what
sets them apart is that data
scientists:
• Bring greater business
acumen to the table
and spend more timethinking strategically
about an organization’s
data resources.
• Sift through all incoming
data with the goal of
discovering a previously
hidden insight, which
in turn can provide a
competitive advantage
or address a pressing
business problem.
• Communicate findings
to both business and
IT leaders in a way that
can influence how an
organization approaches
a business challenge.
http://../jramires/Library/Caches/Adobe%20InDesign/Version%207.5/en_US/InDesign%20ClipboardScrap1.pdfhttp://../jramires/Library/Caches/Adobe%20InDesign/Version%207.5/en_US/InDesign%20ClipboardScrap1.pdf
-
8/15/2019 Ibm Big Data Whitepaper
4/13
ziffdavis.com 4 of 13
Ziff Davis | White Paper | A Visualization is Worth a Thousand Tables
data stores in Hadoop, organizations are often stillrelying on legacy approaches to reporting. Distilling
usable insights from such a wide variety of data is
increasingly difficult, even as users at all levels of
an organization are demanding decision support,
predictive capabilities, real-time data access, and
targeted information.
Only statisticians like tables
Most humans are wired for visual learning and
understanding. Very few people can pour over datatables and pick out trends or grasp big picture issues.
Even statisticians and data scientists who are better
equipped than most to deal with data in all its forms
benefit from interactive graphical interpretations of
data. More importantly, data scientists need to be able
to communicate in straightforward, visual ways with broad, heterogeneous audiences.
Yet while basic graphs and charts are excellent communicators in most cases, creating easily
understood interactive graphics from complex data is remarkably difficult with applications like
Excel. And, as data scales and users require predictive insights, traditional approaches and
tools for data reporting often fail.
Spreadsheets do not equal data visualization tools
Spreadsheet applications are incredibly useful for certain types of calculations, recordkeeping,
tabulation, and many other business tasks. However, they are very poor exploratory and
discovery tools. Despite this limitation, they remain in widespread use in analytical settings.
Similarly, the existing graphical capabilities in spreadsheets and even some major analytics
tools are far too limiting and require too much user input to enable users to efficiently visualize
and explore large or complex datasets. Instead of understanding what the data is telling them
about business trends, users spend their time attempting to create useful pictures.
”How much time do you spend manipulating complicated datasets to generate agraphic that actually communicates something in a slide deck?“
What if you don’t have a data
scientist?
Data scientists are fairly
common in large enterprises,
but many mid-sized businesses
are wrestling with big data
issues without the benefit of
dedicated staff. This is where
user-friendly analytics and
visualization tools come in,
allowing a variety of users to siftthrough and explore data and
succinctly communicate their
insights.
http://../jramires/Library/Caches/Adobe%20InDesign/Version%207.5/en_US/InDesign%20ClipboardScrap1.pdfhttp://../jramires/Library/Caches/Adobe%20InDesign/Version%207.5/en_US/InDesign%20ClipboardScrap1.pdf
-
8/15/2019 Ibm Big Data Whitepaper
5/13
ziffdavis.com 5 of 13
Ziff Davis | White Paper | A Visualization is Worth a Thousand Tables
The Solution: Rich Visualization
Visualization is a natural progression for business analytics, delivering compelling, interactive
and highly descriptive graphics that are well-suited to a variety of data types. Visualization
complements standard written, tabular, and graphical reporting practices, simplifying many
aspects of presenting data and delivering business insights to users. Perhaps a more impor-
tant advantage of visualization, though, is that it enables exploratory analysis that is powerful
enough for data scientists and accessible enough for end users without specific background
in statistics.
What is visualization?
To better understand visualization and ways that it can add value for organizations,
it’s useful to view examples from modern tools generated with different types of
data. While some of the visualizations shown below are fairly typical charts and
graphs, others are state-of-the-art images that take entirely different approaches to
presenting data.
Bar Chart
This stacked bar chart shows a single measure for multiple sales categories over time.
http://../jramires/Library/Caches/Adobe%20InDesign/Version%207.5/en_US/InDesign%20ClipboardScrap1.pdfhttp://../jramires/Library/Caches/Adobe%20InDesign/Version%207.5/en_US/InDesign%20ClipboardScrap1.pdf
-
8/15/2019 Ibm Big Data Whitepaper
6/13
ziffdavis.com 6 of 13
Ziff Davis | White Paper | A Visualization is Worth a Thousand Tables
Radar Chart
A radar chart that shows weekly cycles of sales data that are arranged in a circular
fashion to better illustrate relative fluctuations of data points over time.
Calendar Heat Map
A calendar heat map example shows two years of changes (in percentages) in
customer web orders by year (row), month (column), day of week (sub row), week (sub
column) and day. Heat maps can replace line charts with a more intuitive and compact
representation of layered data.
http://../jramires/Library/Caches/Adobe%20InDesign/Version%207.5/en_US/InDesign%20ClipboardScrap1.pdfhttp://../jramires/Library/Caches/Adobe%20InDesign/Version%207.5/en_US/InDesign%20ClipboardScrap1.pdf
-
8/15/2019 Ibm Big Data Whitepaper
7/13
ziffdavis.com 7 of 13
Ziff Davis | White Paper | A Visualization is Worth a Thousand Tables
Theme River
A theme river is useful for visualizing unstructured and text-based data. This theme river
shows phrase popularity related to gaming platforms over time.
http://../jramires/Library/Caches/Adobe%20InDesign/Version%207.5/en_US/InDesign%20ClipboardScrap1.pdfhttp://../jramires/Library/Caches/Adobe%20InDesign/Version%207.5/en_US/InDesign%20ClipboardScrap1.pdf
-
8/15/2019 Ibm Big Data Whitepaper
8/13
ziffdavis.com 8 of 13
Ziff Davis | White Paper | A Visualization is Worth a Thousand Tables
Social Network Visualization
A social network visualization can show patterns of customer sentiment, key
influencers, and their reach. As with theme river above, this type of graphic would be
impossible to generate with traditional tools like spreadsheets and even many analytical
applications.
http://../jramires/Library/Caches/Adobe%20InDesign/Version%207.5/en_US/InDesign%20ClipboardScrap1.pdfhttp://../jramires/Library/Caches/Adobe%20InDesign/Version%207.5/en_US/InDesign%20ClipboardScrap1.pdf
-
8/15/2019 Ibm Big Data Whitepaper
9/13
ziffdavis.com 9 of 13
Ziff Davis | White Paper | A Visualization is Worth a Thousand Tables
Hierarchy Visualization
A hierarchy visualization represents the relative magnitude of data points, as well
as subsets of data, using bubbles of various sizes and colors. In this example, the
visualization shows the number of targeted campaign responses on regional, state and
city levels. Notice that a single scale is applied across all bubbles, replacing nested or
sequential pie charts with a single graphic.
http://../jramires/Library/Caches/Adobe%20InDesign/Version%207.5/en_US/InDesign%20ClipboardScrap1.pdfhttp://../jramires/Library/Caches/Adobe%20InDesign/Version%207.5/en_US/InDesign%20ClipboardScrap1.pdf
-
8/15/2019 Ibm Big Data Whitepaper
10/13
ziffdavis.com 10 of 13
Ziff Davis | White Paper | A Visualization is Worth a Thousand Tables
Tree Map
Tree maps display data as nested rectangles, the size of which represent their
hierarchical importance in a dataset. This is a tree map view of a social network’s track
selections from a streaming media service.
There are many other types of visualizations that either enhance or replace more
familiar types of charts. Others, like chord diagrams (used to show relationships
among related entities) or the tree map above represent entirely new ways of
looking at data. Regardless of the type of visualization used, the key differentiators
between visualization and standard reporting techniques are interactivity and intuition.
Visualization appeals to our inherent understanding of spatial relationships and
graphical illustrations while enabling us to drill down into data to more fully explore
and examine areas of interest.
Why visualization is different than reporting
Visualization is not a replacement for reporting, whether those reports are delivered in
dashboards, scorecards, or more detailed write-ups describing business issues. Rather,
visualizations are powerful supplements that can be incorporated in reports and presentations
that better communicate complex ideas, relationships, and trends than data tables or more
traditional graphics.
http://../jramires/Library/Caches/Adobe%20InDesign/Version%207.5/en_US/InDesign%20ClipboardScrap1.pdfhttp://../jramires/Library/Caches/Adobe%20InDesign/Version%207.5/en_US/InDesign%20ClipboardScrap1.pdf
-
8/15/2019 Ibm Big Data Whitepaper
11/13
ziffdavis.com 11 of 13
Ziff Davis | White Paper | A Visualization is Worth a Thousand Tables
Even if data visualization incorporates simple bar charts like the first example shown above oruses less sophisticated graphical elements, the presentation should suit the underlying data,
rather than attempting to shoehorn data into limiting legacy graphics. Visualization is about
creating business insight, rather than simply reporting on collected business data.
Exploratory data analysis and predictive analytics – Visually
Perhaps the most powerful use of visualization is in exploratory data analysis. Pivot tables
and other spreadsheet-based tools don’t handle big data well and are quite limiting when
users want to explore relationships in large, complicated datasets. It is far easier to tease out
these relationships with visualizations designed to display temporal, hierarchical, and thematic
associations.
At the same time, users at all levels of a business are trying to derive predictive insights from
these complex datasets. Forrester Research explained how predictive analytics should be
implemented in the enterprise with the following diagram in their 2013 report,
“The Forrester Wave: Big Data Predictive Analytics Solutions, Q1, 2013”:
The tasks of monitoring, understanding, preparing, and evaluating data are all especially
well-served by visualization, particularly when users are working with web-scale and/or
unstructured data.
http://../jramires/Library/Caches/Adobe%20InDesign/Version%207.5/en_US/InDesign%20ClipboardScrap1.pdfhttp://www.ibmbigdatahub.com/whitepaper/forrester-wave-big-data-predictive-analytics-solutions-q1-2013http://www.ibmbigdatahub.com/whitepaper/forrester-wave-big-data-predictive-analytics-solutions-q1-2013http://../jramires/Library/Caches/Adobe%20InDesign/Version%207.5/en_US/InDesign%20ClipboardScrap1.pdf
-
8/15/2019 Ibm Big Data Whitepaper
12/13
ziffdavis.com 12 of 13
Ziff Davis | White Paper | A Visualization is Worth a Thousand Tables
How analytics visualization creates
competitive advantages
If business insights from data analytics are confined
to data scientists and executive leadership, an
organization may benefit strategically, but building real
competitive advantages from data requires insight from
all lines of business. Operations must be data-driven
as well and for that to occur, users need a means
of quickly and easily exploring, understanding and
leveraging an organization’s intricate data resources.
Visualizations are the key to realizing the promise of
big data analytics at both operational and strategic
levels. They are also democratizing, letting bothmid-sized businesses and large enterprises tap into
their data in ways that were previously out of reach for
many users.
IBM Visualization: Natural, Intuitive,and Powerful
IBM Business Analytics solutions include industry-
leading tools for creating the types of visualizations
described in this document. These tools are designed
to support a variety of users, from line of business
decision-makers to statisticians and data scientists
with fewer limitations, less complexity, and more useful
automation than other solutions. IBM’s visualization
solutions meet the needs of mid-sized businesses and
large enterprises, both of which can derive substantial
benefits from the powerful, intuitive, graphical presen-
tation of increasingly complex data.
Visualizations that automatically suit the data
Where spreadsheet-based approaches to creating
charts and graphs require significant user input,
manipulation, and manual selection of appropriate
graph types, IBM Business Analytics tools incorporate
substantial data intelligence and advances in cognitive
computing. This means that IBM SPSS Visualization
The IBM Analytics Ecosystem
IBM has several platforms and
tools that let executives, data
scientists, and line of business
users analyze and, more
importantly, visualize data in
support of business goals and
objectives:
• IBM Cognos Business
Intelligence includes a library
of visualization types that canbe downloaded from IBM
Analytics Zone and customized
for reports. Cognos Business
Intelligence also offers
independent, desktop business
intelligence that features
visualization as a way to better
explore data.
• IBM SPSS Modeler includes
built-in visualization capabilities
to help users better understand
their data mining and models.
• IBM SPSS Analytic Catalyst
creates visualization from big
data and other complex data
sources.
• IBM Cognos Express offers
extensible visualization
capabilities to help midsize
companies and workgroups
better interpret and analyze
data.
• IBM SPSS Visualization
Designer enables the creation
and sharing of customized data
visualizations.
http://../jramires/Library/Caches/Adobe%20InDesign/Version%207.5/en_US/InDesign%20ClipboardScrap1.pdfhttp://../jramires/Library/Caches/Adobe%20InDesign/Version%207.5/en_US/InDesign%20ClipboardScrap1.pdf
-
8/15/2019 Ibm Big Data Whitepaper
13/13
ziffdavis.com 13 of 13
Ziff Davis | White Paper | A Visualization is Worth a Thousand Tables
Designer, for example, can suggest the types of visualizations best suited to the underlyingdata. If a dataset includes specific time series information, the application may suggest a heat
map; they may suggest a chord diagram when data elements have clear directed relationships,
even of these relationships aren’t obvious to the user.
Interactivity
Many of the built-in visualizations available for IBM Cognos Business Intelligence and IBM
SPSS Analytic Catalyst also have interactive elements. Thus, even highly detailed visualizations
can allow users to drill down through specific relationships or isolate subsets that further
enhance their understanding of the data.
Extensibility and ready-made solutions
Finally, IBM offers a library of ready-made visualizations for Cognos Business Intelligence.
In addition, many of the graphics generated in Cognos and SPSS can be customized and
extended, giving organizations the flexibility to present, explore, and act on their data in
whatever ways are the most useful and informative to their users.
Conclusion: Visualization Makes Big Data Accessible
Visualization is not just a logical progression of business analytics, business intelligence,
or even predictive analytics. It is critical to the broader adoption of analytical, data-driven
approaches outside of IT and the offices of data scientists. As BI platforms become more
commonplace, organizations are looking for ways to make these tools truly useful across theirorganizations. For analytics to generate meaningful insights at both executive and operational
levels, organizations need to go beyond standard dashboards and scorecards. Embedding
better pie charts in a slide deck or quarterly report won’t provide widespread benefits or
increase competiveness either.
Instead, delivering on the promise of big data and making analytics broadly accessible to
users and data scientists alike requires fast, interactive and intuitive visualization of data in a
variety of formats. These visualizations should be suited to heterogeneous data and diverse
audiences, providing dynamic views of data and supporting the predictive needs of users with
widely varying interests and areas of expertise. Requirements like these won’t be met with
spreadsheets or even many of the dedicated analytics applications on the market. They will bemet with powerful, intelligent visualization tools built on robust analytics platforms.
To learn more about IBM Business Analytics and visualization, visit:http://www-01.ibm.com/software/analytics/many-eyes/
http://../jramires/Library/Caches/Adobe%20InDesign/Version%207.5/en_US/InDesign%20ClipboardScrap1.pdfhttp://www-01.ibm.com/software/analytics/many-eyes/http://www-01.ibm.com/software/analytics/many-eyes/http://../jramires/Library/Caches/Adobe%20InDesign/Version%207.5/en_US/InDesign%20ClipboardScrap1.pdf